EP3013009B1 - Network assembly and method for detecting a utilisation peak in a convergent network - Google Patents

Description

The invention relates to a method for detecting a utilization extremity in a convergent network, in particular in an industrial network in which data of different services are transmitted with the same or different priorities, wherein a transmitted amount of data from at least one service, in particular at least one transmitting or one is measured during a diagnostic interval, wherein from the measured amount of data of the service and the duration of the diagnostic interval, a utilization value for this service, in particular the bandwidth used, is calculated.

The invention further relates to a corresponding network arrangement as well as network devices which are each suitable for implementing the method according to the invention.

Such a method can be used in many industrial sectors, in particular in the automation and automotive industries as well as in the energy sector. There, the requirement exists for communication networks plant-wide on the basis of a single standardized protocol for all communication services, from the fieldbus level to the business level to be suitable. So there is a converged network where different services and real-time services should be realized. Such networks are also referred to as a convergent time-sensitive network (TSN). By way of example, communication based on the Ethernet protocol can be mentioned here for the industrial sector. Despite the coexistence of different services on the same network, real-time services, such as For example, in the transmission of control data, for each path between the control unit and terminal, predetermined requirements such as minimum delay times (low latency), minimum deviations from the power stroke and minimum Deviations in the duration of data packets (low-jitter) are met.

In order to be able to guarantee for each communication service in the convergent network Quality of Service (QoS), ie, among other things, the fulfillment of the requirements low latency and low jitter, temporary or permanent overloads must be detected early and quickly remedied. This applies both to cyclic data traffic, such as the continuous transmission of control data (including control data streams or CD streams) and audio-video streams (including AV streams) with a respective reserved bandwidth for these services, as well as for The acyclic traffic, the so-called best effort traffic (BE traffic) without reserved bandwidth, which should be transmitted at the speed currently possible for this service, without any restrictions on the other services.

Previously, ingress-policing at the input ports and egress policing at the output ports by known mechanisms such as leaky bucket prevents cyclic data streams, e.g. CD streams or AV streams consume more than the bandwidth reserved for them. If, for example, a data stream has consumed its reserved bandwidth, its telegrams are deleted for a certain period of time.

Information about the actually used bandwidth of a data stream or the BE traffic at a port would be necessary to distinguish between a temporary overload, eg. B. by a data stream or a BE traffic congestion in a network node, and a permanent overload, z. B. by a faulty network node that sends without interruption too much data streams or BE traffic to distinguish.

Out JONGHOON LEE ET AL, "IEEE COMMUNICATIONS MAGAZINE, IEEE SERVICE CENTER, PISCATAWAY, US, (20080401) vol. 46, no. 4, ISSN 0163-6804, pages 160-167 An Ethernet switch architecture is known that allows for identification of subscriber traffic without access to higher level protocols.

Out CARLOS ROJAS ET AL, "Guidelines for Industrial Ethernet Infrastructure Implementation: A Control Engineers Guide", CEMENT INDUSTRY TECHNICAL CONFERENCE, 2010 IEEE-IAS / PCA 52ND, IEEE, PISCATAWAY, NJ, USA, (20100328), ISBN 978-1-4244-6407-4, pages 1-18 It is known that, for diagnostic purposes, the entire traffic is detected via a port and the nature of the traffic is determined separately. The invention is therefore based on the object to be able to recognize and distinguish in a convergent network arrangement utilization extremes.

The invention is further based on the object of providing a network arrangement and corresponding network subscribers who can carry out the method.

This object is achieved by a method for detecting a load extremity in an industrial converged network according to claim 1. In this case, data of different services are transmitted in the convergent network, wherein a transmitted amount of data of at least one service, in particular at least one transmitting or one receiving port, during a diagnostic interval is measured, wherein from the measured amount of data of the service and the duration of the diagnostic interval a utilization value for this Service and this port is calculated, whereby the load values for the respective significant services present in the network are calculated for all transmitting and / or receiving significant ports. The invention is based on the finding that with the methods known from the prior art, detection of overload situations and a distinction between temporary and permanent overload situations is not possible or only too late. The utilization value is a measure of the utilization of the convergent network and can be, for example, a bandwidth used.

The inventive method allows compliance with certain criteria, such as the "low-jitter" and "low-latency" criteria mentioned above.

In this embodiment, services may have the same or different priorities. Significant services are services that, by their nature, quantity, frequency or other characteristics relevant to the flow of data in the network and can, for example, slow down or completely prevent overloading. In a converged network, all services can be significant.

Significant ports are all ports whose load has an influence on the data flow in the network, not all ports in the network have to be significant.

For example, with redundant networks, the redundant ports can be defined as insignificant in order to save computing power and energy. If a port fails and the redundancy has to step in, this port can again become significant. A port can also be significant if, for example, more than 60% of the traffic is handled by it.

Thus, in a preferred embodiment, all ports and all services may be significant, ie, for each service, the actual bandwidth consumed, for example the number of bytes per second, is measured at each transmitting or receiving port. For this purpose, the number of received bytes is counted within adjustable time intervals at each receive port for each data stream and for the BE traffic. Analogously, the number of bytes sent is counted within adjustable time intervals at each send port for each data stream and for the BE traffic. Here, for example, the start times and the duration of the diagnostic intervals with an accuracy of z. B. 1 ns for each transmitting or receiving port and independently adjustable for each service. Finally, all utilization values can then be calculated and processed further. It is particularly advantageous that constantly the necessary data for the detection of utilization extremes are available. According to the invention, the utilization value of a service is compared with a first threshold value, and when it is exceeded, a temporary overload is detected. This first threshold value can be permanently stored in the memory of the port of the device itself, for example. Also possible in the network device are integrated memory, central memory like a server, with connection to the network, decentralized memory for one group each Network devices, e.g. As distributors such as switches and hubs, etc. It is advantageous that a temporary overload is detected very quickly and appropriate countermeasures can be initiated.

Further advantageous embodiments make it possible to compare the utilization value of a service with a storage value of the maximum utilization for this service and / or a further storage value of the global maximum utilization, and the current utilization value is stored as the respective new maximum utilization upon reaching or exceeding the respective stored maximum utilization. Memory values can be stored in the already mentioned memories. This has the advantage that the stored maximum utilization value is always available for further evaluations.

Analogously, the calculated utilization value of a service can be compared with a storage value of the minimum utilization for this service and / or another storage value of the global minimum utilization, and the current utilization value can be stored as the respective new minimum utilization upon reaching or falling below the respective stored minimum utilization. Furthermore, according to the invention, the sum of the measured data volume of a service measured during the diagnostic intervals is formed and an average of the network utilization is calculated with the sum of the duration of all the diagnostic intervals. On the basis of this average, first indications can be identified, which may, for example, indicate a low utilization of the network. The average of the network load will also be compared to a second threshold to detect permanent network congestion if exceeded. This has the advantage of being able to search directly for the cause of this persistent overload, for example, to network subscribers incorrectly in large quantities to send data to identify and put back into a regular operating mode.

As an advantageous embodiment, it can also be considered that the various threshold values and the local and / or global maximum / minimum utilization values can be adapted dynamically by a network user or a higher-level entity. This offers the great advantage that the threshold values can be adapted situationally. For example, during the night shift or the weekend, when AV traffic is reduced to a minimum by the absence of most employees, more bandwidth can be freed up for other services or data types. In this way, large amounts of data can be transferred from backups or complex production data without triggering overload detection.

The further object of the invention is achieved by a network arrangement for carrying out the method described. This network arrangement can, for example, via Ethernet; Wireless via WLAN or Bluetooth; be connected via PowerLAN or other known network systems or fieldbus systems.

Particularly advantageous is an embodiment in which the network arrangement part of an industrial automation system or can be coupled thereto. Similarly, a solution is possible in which this network arrangement is used in a cyber-physical system and thus enables Industry 4.0-compliant production in which not only the production is networked, but also the engineering and even the sourcing is linked. Thus, for example, small series can be produced more efficiently.

Another object of the invention is advantageously to provide various network devices which are suitable for carrying out the method and for integration into the network arrangement. Among others, there are switches, Hubs, routers, network adapters or other network distributors and interfaces. Industrial controllers with corresponding ports as well as other industrial control and switching applications can also be used here.

With the present embodiments, it is now possible to distinguish whether temporary or permanent overload exists, and to make a statement for each significant receive or transmit port and for each significant service. Thus, it is also possible to monitor several services, in particular best-effort, control-data or audio-video services. The detection or distinction of a temporary or permanent overload in cyclic traffic (CD streams, AV streams) or acyclic traffic (BE traffic) in a converged network allows fast and effective countermeasures such. B. in a permanent overload situation when receiving a data stream discarding the affected data stream. The fast detection of an overload and a consequent fast response facilitates the use of various services such as CD streams, AV streams, BE traffic in a converged network in the industrial sector. The ability to deploy these disparate services on the same network not only increases the power of industrial communications, but also reduces network cost and time to market by reducing the number of network components required for a converged network compared to individual service networks are.

FIG 1

eine konvergente Netzwerkanordnung,

FIG 2

einen Arbeitstakt in einer konvergenten Netzwerkanordnung,

FIG 3

einen beispielhaften Datenstrom in einer konvergenten Netzwerkanordnung.

In the following the invention will be described and explained in more detail with reference to the embodiments illustrated in the figures. Show it:

FIG. 1

a convergent network arrangement,

FIG. 2

a working clock in a convergent network arrangement,

FIG. 3

an exemplary data stream in a convergent network arrangement.

The FIG. 1 1 shows an embodiment of a convergent network arrangement 1 which consists of a plurality of different network subscribers 30, 32, 34 and 36. A central network distributor 30 is connected through its ports 12a and 12b to a redundantly connected network device 32 via a network connection 2 and a redundant network connection 3 through its ports 12e and 12f. The redundant network connection 3 is only activated when the network connection 2 no longer functions properly. As long as the network connection 2 is functioning properly, the associated ports 12a and 12e can be significant and a method according to the invention can be carried out. In the event of an error, the network connection 2 is deactivated, the ports 12a and 12e are no longer significant, the redundant network connection 3 is activated and the ports 12b and 12f become significant. Furthermore, the central network distributor 30 is connected by port 12d to another network distributor 34 via a network connection 2 through its port 12g. Port 12c is not used and is therefore not significant. Memory values 14 and threshold values 16 are stored by way of example in a memory 13 of the central network distributor 30. It would also be conceivable that each of the network devices 30, 32, 34 and 36 has its own memory 13 or, alternatively, all other network distributors 34 have a memory 13.

The network terminals 36, for example PCs, industrial interface operating interfaces, industrial controllers or other network-capable devices, are each connected via a network connection 2 to the ports 12h and 12i of the network distributor 34. Whether these ports and the data sent and / or received about them are significant depends on the application. For example, it may be a terminal that plays an important role in production and the data volume is very high, it is recommended to define the involved ports as significant ports.

The network arrangement 1 does not have to be realized physically but can be implemented partially or completely virtualized.

The FIG. 2 shows an example of a working clock 7 of the convergent network arrangement 1 in which the maximum amount of data is transmitted. The data blocks 4 represent the smallest possible unit in which data can be transmitted, for example 1 byte. Depending on the network, these smallest possible data blocks 4 may also be larger or smaller than one byte. In this example, no service A, B, or C has yet been assigned to the data blocks 4. The services are, depending on the respective network and its field of application, correspondingly diverse and not limited to A, B, or C.

The FIG. 3 shows by way of example the data transmission at a transmitting or receiving port 12. In this case, a working clock 7 is the smallest possible interval during which data can be transmitted. Each data block 4 is assigned a service A, B, C, which in turn decides whether this service is significant and thus should be detected by the method according to the invention. Also visible is the diagnostic interval 6, which in this case extends over three working cycles, but can also assume any further multiples of the working cycle. It is, for example, the amount of data, ie the sum of all data blocks 4 of the service A, which are in the diagnostic cycle 6 during the power strokes 7 counted then from the duration of the diagnostic interval 6 a utilization value, beispielswese the bandwidth to calculate.

In summary, the invention relates to a method for detecting a utilization extremes in a convergent network, in particular in an industrial network and / or a "time-sensitive network" (TSN), in which different data Services with the same or different priorities. A transmitted amount of data of at least one service is measured in particular on at least one transmitting or one receiving port during a diagnostic interval. From the measured data volume of the service and the duration of the diagnostic interval, a utilization value for this service and this port is calculated. The load values for the relevant significant services present in the network are calculated for all sending and / or receiving significant ports. A corresponding network arrangement for carrying out the method is likewise part of the invention. The invention has for its object to be able to recognize and distinguish in a convergent network arrangement utilization extremes.

Claims (7)

1. Method for detecting a utilisation peak in an industrial convergent time-sensitive network (1), in which data (4) from various services (A, B, C) is transmitted, wherein at least one of the services (A, B, C) transmits control data and a further of the services (A, B, C) transmits acyclic data in the form of best effort traffic, wherein a transmitted data quantity of at least of one of the services (A, B, C) is measured on at least one sending or one receiving port (12) during a diagnosis interval (6),
   characterised in that
   a utilisation value for the respective service (A, B, C) and this port (12) are calculated in each case from the measured data quantity of the respective service (A, B, C) and the duration of the diagnosis interval (6) in each case and wherein the utilisation values for the respective significant services (A, B, C) present in the network are calculated for all sending and/or receiving significant ports (12),
   wherein significant services (A, B, C) are those services which, on account of their condition, quantity, frequency or other properties, are relevant to a data flow in the network, wherein significant ports (12) are those ports whose utilisation has an influence on the data flow in the network, wherein the utilisation value of a significant service (A, B, C) per significant port (12) is compared with a first threshold value (16) and when the same is exceeded, a temporary overload is detected, wherein per significant port (12), a sum of the transmitted data quantity of the respective significant service (A, B, C) measured during the diagnosis interval (6) is formed for all significant ports (12) and with a sum of the duration of all diagnosis intervals (6) an average value of the network utilisation for the respective significant services (A, B, C) present in the network is detected,
   wherein the average value of the network utilisation is compared with a second threshold value (18) and when a permanent network overload is exceeded is detected by the respective significant service.
2. Method according to one of the preceding claims, characterised in that the utilisation value of a service (A, B, C) is compared with a storage value (14) of the maximum utilisation for this service and/or a further storage value (14) of the global maximum utilisation and when the respectively stored maximum utilisation is reached or exceeded, the current utilisation value is stored as the respective new maximum utilisation.
3. Method according to one of the preceding claims, characterised in that the calculated utilisation value of a service is compared with a storage value (14) of the minimal utilisation for this service and/or a further storage value (14) of the global minimal utilisation and when the respectively stored minimal utilisation is reached or exceeded, the current utilisation value is stored as the respective new minimal utilisation.
4. Method according to one of the preceding claims, characterised in that the various threshold values (16, 18) can be adjusted dynamically by a network subscriber (30, 32, 34, 36) or a higher-order entity.
5. Network device for carrying out a method according to one of claims 1 to 4, in particular switches, hubs, routers, network adapters or further network distributors.
6. Network assembly with at least one network device according to claim 5.
7. Network assembly according to claim 6, characterised in that the network assembly is part of an industrial automation system or can be coupled thereto.

Images